Extended Data Fig. 9: Computational model.
From: Antibody and TLR7 agonist delay viral rebound in SHIV-infected monkeys
a, LASSO and PLSR model identifies the parameters that correlate with delayed viral rebound (n = 11 monkeys per group). Left, individual monkeys are shown distributed by latent variables 1 and 3 of the model. Timing of viral rebound is indicated by the colour gradient. R2 = 0.176, root mean square error (RMSE) = 0.917, P < 0.001 in two-sided permutation tests. Middle, the contribution of the selected features to model separation is displayed in variable importance in projection (VIP) scores, related to early (blue) or late (red) viral rebound. Right, correlation between viral rebound and latent variable 1. P value reflects a two-sided Spearman rank-correlation test. b, LASSO and PLSR model identifies the parameters that correlate with reduced total viral loads (n = 11 monkeys per group). Left, individual monkeys are showed distributed by latent variables 1 and 2. Total viral loads are indicated by the colour gradient. R2 = 0.282, root mean square error (RMSE) = 0.857, P < 0.001 in in two-sided permutation tests. Middle, the contribution of the selected features to model separation is displayed in VIP scores, related to high (blue) or low (red) total viral loads. Right, correlation between total viral loads and latent variable 1. P value reflects a two-sided Spearman rank-correlation test.
